An important landmark in scientific publishing
In this volume Gaillard, Karten, and Sauvé, in their article “Retinorecipient areas in the diurnal murine rodent Arvicanthis niloticus: A disproportionally large superior colliculus” (Gaillard et al., 2013) achieve a milestone in scientific publishing. In support of their findings, they provide an unprecedented level of access to their supporting data by publishing their full set of experimental outcomes in the form of virtual slides, or whole-slide images. Using the image streaming technology, Biolucida Cloud from MBF Bioscience, Gaillard et al. have placed their entire experimental image collection for this study on a publicly accessible server at http://wiley.biolucida.net:80/link?l=FTtbQt. This collection totals over forty gigabytes in its compressed form and would utilize hundreds of gigabytes if not for modern compression methods. This example of publishing comprehensive results implements a completeness that we believe will result in more informed peer review, a finer granularity of interpretation at the reader level, the promotion of related research since other anatomic regions are fully scanned, and longevity of original image data.
In comparative anatomic studies, image data often comprise a large portion of experimental observation. Due to technologic limitations, researchers currently publish only a fraction of the image data acquired during an investigation. Further, what is published has limited levels of pixel resolution, with authors relying on figure insets to display both high and low magnification views of their specimens. These constraints limit the data available to the reader, and provide only a limited window on the experimental material. The reader is highly reliant upon the interpretations and scientific perspective of the author. First, they prevent reviewers and readers from making fully informed judgments about data quality. Interpretation is limited to conclusions drawn from a static published subset of a study's data. Second, limits in the amount and resolution of published image data restrict interpretation and evaluation of specimens to specific anatomic areas. This makes overall assessment of preparation quality difficult and makes examinations of other anatomic regions at full resolution impossible. Finally, physical slides require physical storage space and significant effort to create an efficient, scalable, and discernible organizational schema. Storage for physical slides typically spans shelving and freezers and utilizes laboratory-specific organization that may lose coherency with changes in laboratory members. Additionally, many slides, particularly fluorescent preparations, have a limited lifetime.
Collections of serial sections of brains of diverse animals has long been an important part of neuroanatomy, and is an essential practice in comparative and evolutionary neurobiology, as evident in the collections of Santiago Ramón y Cajal, Lüdwig Edinger, Charles Judson Herrick, Cornelius U. Arlëns-Kappers, Elizabeth Crosby, and others. Sharing these precious collections with scientists in other laboratories was extremely difficult, often requiring significant travel time and expense. Furthermore, many such collections have suffered deterioration with the passage of time. The collections are vulnerable to breakage of the fragile glass slides, and many valuable collections were lost with the retirement or death of the collectors. In an attempt to deal with these problems, John Irwin (Jack) Johnson and Wally Welker organized one of the first efforts to assemble a library of images of various mammalian brains, and share them using the capabilities of the internet (http://www.brainmuseum.org). The technology available at that time, however, could only produce low-resolution images of complete sections of brains, and the time and labor involved often limited the number of sections available from any single brain.
The most recent major advance occurred in about 2007 when the late Edward (Ted) G. Jones obtained a new generation slide scanner from Aperio. Originally developed for quickly sharing histological images in Pathology labs, whole slide imaging has emerged as a routine procedure in pathology and histology research. Ted Jones and Shawn Mikula realized that this scanner was capable of scanning a complete section of a monkey brain in RGB color, in 15 minutes, at a resolution of 0.5 μm/pixel. Ted Jones, Shawn Mikula, and Jim Stone revolutionized neuroanatomy with their development of a massive database of brain images, instantaneously available over the internet to labs around the world (http://www.brainmaps.org) (Jones et al., 2011). Combined with the falling costs of data storage, increased speed of the internet, and improvements in slide scanners, they demonstrated how this technology had the potential to transform experimental neuroanatomy as well as sharing rare and precious brain collections. Ted Jones demonstrated the priceless value of this technology with his posting of complete sets of serial sections of brains of duck-billed platypus (Ornithorhynchus anatinus) and echidna or spiny anteater (Tachyglossus aculeatus) as well as such exquisite datasets as 740 serial sections of the grivet monkey (Chlorocebus aethiops), all at 0.5 μm/pixel in full color. Further demonstration of the power of this technology was evident in serial section data sets in all three ordinal axes of retinal projections in mice, chicks and goldfish. Individual axons could easily be visualized and their density and distribution to various regions of the brain independently evaluated by neuroanatomists around the world. Similar reference data sets showing the distribution of tyrosine hydroxylase, choline acetyltransferase, parvalbumin, calbindin, are now available to scientists at sites such as: http://www.brainmaps.org, http://www.zebrafinch.org/anatomy, as well as from the Allen Institute (http://www.alleninstitute.org/), http://www.mouseconnectome.org from UCLA, and http://brainarchitecture.org/mouse from Cold Spring Harbor Laboratory.
TECHNOLOGIC ACHIEVEMENT AND FUTURE WORK
The virtual slide collection accompanying the published work of Gaillard et al. is a remarkable technical feat. To put this in a tangible perspective, a typical virtual slide in the collection would require over 250 square meters of paper if printed at full resolution. Powered by Biolucida Cloud serving technology, these slides can be explored at low and high resolutions, including their native scan resolution (0.5 μm/pixel), from any computer with an internet connection.
In addition to publishing the imagery from their study in its entirety, the authors will also be able to use collaborative technology within Biolucida to indicate to reviewers and readers the exact regions of brain they analyzed. Annotations, consisting of contours, arrows, and text indicate areas of interest for each slide, while bookmarks highlight anatomic areas of interest to the study. These annotations and bookmarks are available to anyone viewing the slides. Together with the unprecedented access to the study's full-resolution imagery, these collaborative tools will allow readers to view the exact slides and anatomic regions used by the authors and will hopefully elucidate the path taken by the authors to reach their conclusions.
One of the most immediate consequences of this technology is that a single set of slides can be prepared and high-resolution images of all the slides can be shared among collaborators around the globe. The study by Gaillard et al. in the current issue of the Journal of Comparative Neurology involved evaluation of the data by authors in France, United States, and Canada. The authors were able to evaluate remotely the exact same image in order to compare their interpretation of the data. Serial digital images allowed precise section-to-section analysis of nuclear organization and patterns of termination of retinal axons. Complete datasets of the retinal projections of brains cut in all three major ordinal planes (transverse, sagittal, and horizontal) allowed instantaneous comparison of axon trajectories and nuclear organization in multiple planes, facilitating definition of the boundaries of cell groups.
Other researchers can also utilize this comprehensive collection to examine other anatomic areas of brain not considered in the study at hand, due to unintentional spread of tracer within the orbit, with subsequent retrograde labeling of oculomotor neurons and anterograde labeling of trigeminal sensory axons. Biolucida can be used to to pan throughout brain sections at selectable zoom levels, examining anatomies and features relevant to their own areas of investigation. This would allow investigation of other anatomic regions without rescanning slides, or the need to reproduce animal studies.
Finally, the publishing of the image collection at http://wiley.biolucida.net:80/link?l=FTtbQt represents a permanently accessible and easily accessible repository for imagery associated with a study, and potentially for a laboratory or facility. Images and experiments can be organized into collections, tagged with key words and descriptions for later indexing, and annotated for documentation and collaboration.
It is our sincere hope that this example of completeness, transparency, and clear organization, will allow readers to have the most complete understanding of the study possible and encourage the community at large to extend such completeness and openness to the readers of their own published work.
Gaillard F, Karten HJ, Sauvé Y. 2013. Retinorecipient areas in the diurnal murine rodent Arvicanthis niloticus: A disproportionally large superior colliculus. J Comp Neurol doi: 10.1002/cne.23303.
Jones EG, Stone JM, Karten HJ. 2011. High-resolution digital brain atlases: a Hubble telescope for the brain. Ann N Y Acad Sci 1225 Suppl 1:E147-E159.